topic Re: Determining if Random Effects are Significant in Proc HPMIXED in SAS Programming

Determining if Random Effects are Significant in Proc HPMIXED

Ameurgen — Mon, 17 Apr 2023 15:17:43 GMT

Hi every One,

I need your help, i am using mixed model in sas with proc HPMIXEd and i need to loook closley to my random effect if it is significant or not, of course in table of COVPARMS, like fixed effect table analysis and after many reseach hard reseach i cant find any thing clear especially when it comes to proc Hpmixed .

so please Don't skimp on me if you have any information .

Thank you

Re: Determining if Random Effects are Significant in Proc HPMIXED

sbxkoenk — Mon, 17 Apr 2023 15:33:54 GMT

Hello,

On the MODEL statement and on the RANDOM statement , put an "s" (for solution) as an option behind the forward slash.

You will then get Solution for Fixed Effects and Solution for Random Effects. Both estimates and p-values !

Like here :

   model Height = Gender / s;
   random Family Family*Gender / s;

Koen

Re: Determining if Random Effects are Significant in Proc HPMIXED

Ameurgen — Mon, 17 Apr 2023 15:57:44 GMT

Thank you sir,

I appreciate your reply and i have a question this s option after Random statement its will give us the solution of each class of the random effect, i shar with you my covparms estimate , the question is , is 0.17 from HY VARIABLE is significant or not and which test can confim this results.

Thank you again

Covariance ParameterEstimates Cov Parm Estimate

HY =0.1758

dam_YOB = 0.007

Residual =2.7753

Re: Determining if Random Effects are Significant in Proc HPMIXED

sbxkoenk — Tue, 18 Apr 2023 08:06:32 GMT

Hello,

Are you complaining that PROC HPMIXED does not have a COVTEST option and PROC MIXED does? It would have helped to specify that from the beginning of your question. 😉
So I am experiencing it now too (was unaware of it).
Perhaps you could switch to PROC LMIXED (SAS VIYA). That is even faster than PROC HPMIXED.

Cheers,

Koen

Re: Determining if Random Effects are Significant in Proc HPMIXED

Ameurgen — Mon, 24 Apr 2023 16:02:00 GMT

Following my question and since no one can suggest me a solution, i find such good solution to determine wether or not the random effetc is significant or not in the mixed model using Hpmixed, so all we know we can do it with proc mixed since we have the COVTEST statement to meseare this significativity but in such case of big data like i use the option of proc mixed is not suitable required a lot memory s, briefly my idea is using fist hpixed to output your covariance parameter table and after using to next step with proc mixed but here you can added as pdata in PARMS statement and you will get your significant wald test for your random effect. unfortunatly with this i sill not get my results. so i thought to shar this information that will be good for sas communities.

thank you all

AMEUR

Re: Determining if Random Effects are Significant in Proc HPMIXED

sbxkoenk — Mon, 24 Apr 2023 20:29:51 GMT

@Ameurgen wrote:

unfortunately with this I still not get my results.

How do you mean?
Are you getting an out-of-memory ERROR?
Do you stop the program from running after 30 hours (because you are tired of waiting)?

The below paper provides recommendations for circumventing memory problems and reducing execution times for your mixed modeling analyses.

Paper 332-2012
Tips and Strategies for Mixed Modeling with
SAS/STAT® Procedures
Kathleen Kiernan, Jill Tao, and Phil Gibbs, SAS Institute Inc., Cary, NC, USA
https://support.sas.com/resources/papers/proceedings12/332-2012.pdf

Koen